Rancher Longhorn Engine Update

Today my Ghost container wouldn’t start. In the logs I could see that the volume wasn’t available. Rancher Longhorn Engine Update!

What can happen when a Rancher Longhorn engine update takes place? What should you watch out for? That’s what this short post is about.

Engine Update

What does this mean in concrete terms? Longhorn is a still-in-development Kubernetes block storage system (current version at the time of this post is 0.6.1 with engine version 0.6.0). I’ve mentioned Longhorn several times – including in the Kubernetes einfach mit Rancher: Storage article.

Today I noticed that my popel instance on my Kubernetes cluster wasn’t running. Looking at my Ghost workload immediately showed me the problem: the Ghost volume – which is based on Rancher Longhorn – couldn’t be loaded/mounted.

In the Longhorn UI everything looked fine. Last Friday I had updated the installation – via Helm chart. The appearance of my problem and upgrading Longhorn were therefore connected – that was my assumption.

In the volume detail view, I also saw that the replicas weren’t running. An engine update had taken place, from 0.5 to 0.6. Without further research, I did the following – since this is only a test environment, any form of data loss would be acceptable – I tried to update the engine.

Longhorn Engine Update

I first had to find the point – the approach – for the update. The upgrade can apparently only be executed at the volume level.

The upgrade can be executed by clicking the hamburger menu in the volume detail view.

The upgrade can also be started from the volume listing – the upgrade option is also hidden in the hamburger menu there.

This action takes about 2 minutes for my 2 GB large replicas. This process created three new replicas and deleted the existing – old – replicas.

After that, my Ghost workload was able to mount the volume again and start without any problems.